Characterizing Medial Temporal Lobe Neurodegeneration Due To Tau Pathology In Alzheimer's Disease Using Postmortem Imaging

Thumbnail Image
Degree type
Doctor of Philosophy (PhD)
Graduate group
Alzheimer's Disease
Image Registration
Image Segmentation
Computer Sciences
Neuroscience and Neurobiology
Grant number
Copyright date
Related resources
Ravikumar, Sadhana

Tau neurofibrillary tangle (NFT) pathology in the medial temporal lobe (MTL) is closely linked to neurodegeneration, and is the early pathological change associated with Alzheimer’s Disease (AD). Using in vivo structural magnetic resonance image (MRI) analysis, volumetric change in the MTL caused by neurodegeneration can be measured. While these measurements have been shown to be sensitive to structural change during the early stages of AD, the potential of these measures as imaging biomarkers for early AD is limited by the fact that other neurodegenerative pathologies also affect MTL morphometry. Studies have suggested that different subregions of the MTL are differentially affected by AD and non-AD pathologies. Therefore, by characterizing MTL neurodegeneration due to the different pathologies, patterns of structural change specific to NFT pathology can be isolated. In this work, we hypothesize that MRI measurements obtained from MTL regions where structural change correlates most strongly with AD-specific pathology would result in biomarkers that are more sensitive to longitudinal change in the presence of preclinical AD. To elucidate patterns of structural change in the MTL specifically associated with NFT pathology, the work presented in this dissertation leverages a postmortem imaging dataset which combines ultra-high resolution ex vivo MRI with ground truth information from serial histology, thus providing a direct link between structural changes in the MTL and the underlying pathology. Using ex vivo MRI and pathology datasets from a large number of MTL autopsy specimens, this dissertation focuses on the development and validation of computational image analysis techniques for groupwise analysis of the MTL. This work leverages advanced computational anatomy techniques and prior knowledge of cortical geometry to overcome challenges due to the complex and highly variable geometry of the hippocampus and MTL cortex. First, I present novel semi-automated and automated segmentation techniques that facilitate labeling of the MTL cortex and the stratum radiatum lacunosum moleculare (SRLM) layer of the hippocampus in high-resolution ex vivo MRI datasets. Segmentations of these structures are used to guide the alignment of the MTL across different specimens. Groupwise analysis relies on the creation of a normalized reference space across specimens via groupwise image registration. To this end, I explore two approaches for groupwise registration: 1) a volumetric approach which builds on a previously developed shape and intensity-based framework to construct a 3D probabilistic atlas of the MTL, and 2) a surface-based approach which provides implicit registration between specimens by applying a topological unfolding framework to the MTL. As part of this work, we present a first-of-its-kind 3D computational atlas of the MTL, with anatomical subregion labels derived from serial histology on the basis of cytoarchitecture. Leveraging the developed registration frameworks and serial histology datasets, I investigate the relationship between tau pathology and regional MTL cortical thickness, while correcting for age and the presence of co-pathologies, and identify regional patterns of cortical thinning consistent with early histological studies. These analyses are performed using semi-quantitative ratings of neuropathology and quantitative 3D maps of tau NFT burden, and provide a more refined understanding of how tau pathology is associated with neurodegeneration in the MTL.

Paul A. Yushkevich
Date of degree
Date Range for Data Collection (Start Date)
Date Range for Data Collection (End Date)
Digital Object Identifier
Series name and number
Volume number
Issue number
Publisher DOI
Journal Issue
Recommended citation